Deep Learning Kernel Software Performance Architect at NVIDIA

Department: Architect

Language

Setup: On-site
Location: Huangpu, Shanghai · Beijing
Type: Full-time
Level: senior
Posted: Jul 9, 2026

Description

NVIDIA is hiring Software Performance Architects to optimize GPU kernel performance for data-center platforms and deep learning workloads. The role involves performance analysis, debugging, profiling, root-cause isolation, and optimization; building Python-based automation and regression workflows; and collaborating with kernel developers, compiler, architecture, SWQA, and infrastructure teams. Candidates need a master's or PhD in a relevant field, strong Python and C/C++ skills, computer architecture and parallel-programming fundamentals, and experience with performance analysis and regression workflows.

Required Skills

Python
C/C++
Computer architecture
Parallel programming
Profiling
Regression triage
CUDA

Additional Information

Minimum Education: Master's or PhD degree or equivalent experience in Computer Science, Computer Engineering, Applied Math, or a related field
Required Previous Experiences: At least 2 years of experience reading and debugging performance-oriented code
Experience with performance analysis workflows, including profiling, measurement methodology, reproducibility, and regression triage
Preferred Qualifications: Experience with high-performance kernels or math libraries such as GEMM, attention, or CUTLASS-like concepts
GPU programming or performance experience with CUDA or equivalent parallel programming
Strong understanding of machine learning and deep learning workloads, including training and inference shapes, precision modes, and performance bottlenecks
Familiarity with simulators, analytical modeling, or performance characterization methodology
Translation Company Name: NVIDIA